Testing and Analysis of the HRV Signals from Wearable Smart HRV Sensors

Abstract

The objective of the test procedure is to obtain bio signals from Photoplethysmograph and Electrocardiograph sensors on selected consumer devices and to statistically validate the data for use with a drowsiness estimation method.The method selected for validation uses LF/HF ratio calculated by a set of R-R interval data to estimate drowsiness state of a human. The value LF to HF ratio calculates balance between sympathetic and parasympathetic activity that can be measured from HRV (Heart rate variability) signals. The statistical data collected are processed by using Fast Fourier Transform and HRV frequency domain analysis on a set of test participants.There is a correlation between medical ECG equipment control output and Matlab tool’s HRVAS (Burg) output of data processed from ECG based wearable smart sensor when the LF/HF ratio is calculated in all observed volunteer data. The results for Photoplethysmograph sensors of this test correlate with other tested tools but level of the values is lower, and data from optical biosensor devices which are designed to measure HRV time-domain properties as pulse did not confirm with ECG equipment results for frequency-domain analysis required for use with selected drowsiness estimation method. The result affecting factors are sensor placement, motion artefacts and discrete vendor-specific signal pre-processing of wearable device output data.The following results confirm the use of consumer grade biosensor that produces discretely pre-processed R-R interval data for the frequency based HRV method and application validation against directly processed ECG data from certified medical equipment.

Human Emotion Recognition: Review of Sensors and Methods

“…Typical HRV measurements taken from frequency domain analysis are powers within frequency bands and ratios of powers. The amount of power contained within a frequency band can be obtained by integrating the PSD within the band frequency limits [103].…”

Assessing the Accuracy of Popular Commercial Technologies That Measure Resting Heart Rate and Heart Rate Variability

“…HRV, derived from the inter-beat intervals spanning across consecutive heart beats (Shaffer and Ginsberg, 2017 ), provides a more in-depth reflection of auto-regularity modulation within the human body (Acharya et al, 2006 ; Markovics et al, 2018 ). Standard HRV metrics comprise the time and frequency domains, which oscillate in response to an individual's immediate psychophysiological response to events or stressors (Shaffer and Ginsberg, 2017 ).…”

“…To date, HRV metric validation efforts are limited to examining device accuracy during bouts of physical activity (Hernando et al, 2016 ; Bunn et al, 2018 ; Henriksen et al, 2018 ), which is likely a result of the growing interest in HRV monitoring for stress adaptations as they relate to sport performance and physical health (Jiménez-Morgan and Mora, 2017 ). Thorough third-party validations are further limited by manufacturers often incorporating proprietary noise cleaning algorithms (De Arriba-Pérez et al, 2016 ; Henriksen et al, 2018 ; Markovics et al, 2018 ; Bent et al, 2020 ) to allegedly bolster signal quality, as many popular wearable devices are commonly sensitive to motion artifact (Lee and Zhang, 2003 ; Yousefi et al, 2013 ; Waugh et al, 2018 ). These algorithms are rarely disseminated to consumers or tested by independent researchers (De Arriba-Pérez et al, 2016 ).…”

Feasibility study of physiological parameter registration sensors for non-intrusive human fatigue detection system

“…The next stage of sensor evaluation is based on the experimental results, where the sensitivity of the measured parameters and the sensor signal quality can be statistically evaluated by control methods of voluntary measurements and tests that are specific to the problem domain. The sensor comparison and measurements by following a measurement protocol are reflected in the authors' previous published work [29].…”